Marker for predicting insulin fortified treatment effect and application thereof

Abstract

The invention discloses a marker for predicting an insulin fortified treatment effect and application of the marker, and relates to the technical field of biomarkers. In order to solve the problem that in the prior art, there is no research on systematic exploration of the effect of microRNA in SIIT curative effect prediction, it is found through systematic research that hsa-miR-502-3p is an independent predictive factor for SIIT long-term remission, a novel combined prediction model for SIIT long-term remission based on hsa-miR-502-3p and ISSI-2 is constructed, and the model is used for predicting the effect of microRNA on SIIT long-term remission. Compared with a single hsa-miR-502-3p prediction model and a single ISSI-2 prediction model, the prediction model has the advantages that the recognition capability of the model is further improved, and better balance between prediction efficiency and clinical interpretability is realized; in addition, the invention also clarifies the regulation and control relationship between hsa-miR-502-3p and SUR1, reveals a partial molecular biological mechanism of SIIT treatment heterogeneity, and provides a new view angle and basis for T2DM research and SIIT clinical practice.

Description

Technical Field

[0001] The present invention relates to the technical field of biomarkers, and in particular to a marker for predicting the effect of intensive insulin therapy and its application. Background Art

[0002] Diabetes mellitus (DM) is a major challenge in the field of global public health, and its disease burden continues to increase. Among them, type 2 diabetes mellitus (T2DM) is a metabolic disease caused by the interaction of genetic and environmental factors. Its pathophysiological mechanism is complex and has not been fully elucidated. It mainly involves two core factors: insulin resistance and pancreatic β-cell dysfunction. Specifically, insulin resistance is manifested as a weakened response to insulin in metabolic organs and tissues such as fat, muscle and liver, and glucose cannot be effectively used by the human body; β-cell dysfunction is manifested as reduced insulin secretion or abnormal pattern, which cannot meet the body's demand for insulin. Among newly diagnosed T2DM patients, about 50% of patients have a significant hyperglycemia state with glycosylated hemoglobin (HbA1c)>9%. At present, the treatment path of T2DM is mainly to start with a single oral medication, and then use different types of oral medications or insulin in combination with the progression of the disease. However, the blood sugar control rate in my country is still less than 50%. Long-term exposure to high blood sugar leads to a significant increase in the risk of microvascular and macrovascular complications, which has imposed a heavy economic burden on society and seriously affected the quality of life of patients.

[0003] In recent years, studies have shown that it is possible to reverse diabetes through active management, including lifestyle weight loss, metabolic surgery, and short-term intensive insulin therapy (SIIT). SIIT is a treatment method that uses multiple subcutaneous insulin injections or multiple insulin injections per day on the basis of lifestyle intervention and patient education to optimize the control or even normalize the patient's blood sugar level. A number of randomized controlled trials have confirmed that SIIT has excellent clinical value in the early intervention of type 2 diabetes. With the accumulation of clinical follow-up data and evidence-based medicine, the clinical application of SIIT has formed a relatively complete system, and has derived multiple optimized implementation paths such as the intensive-simplified treatment strategy (80.4% of newly diagnosed T2DM patients with an average HbA1c of up to 11.0% achieved a 1-year HbA1c < 7.0% standard rate) and the new short-term intensive insulin therapy (N-SIIT). The long-term drug-free relief of diabetes brought about by SIIT has not only significantly improved the quality of life of patients, but also greatly reduced the public health and medical burden on society. Currently, my country's guidelines have included SIIT as the preferred treatment option for newly diagnosed T2DM patients with severe hyperglycemia symptoms (FPG ≥ 11.1 mmol / L or HbA1c ≥ 9%).

[0004] The attenuation of the benefits after SIIT, that is, the proportion of patients maintaining remission gradually decreases with the increase of follow-up time, highlighting the importance of predicting the efficacy after SIIT. Therefore, constructing an early clinical prediction model to identify and predict the possible efficacy responses of different patients after receiving SIIT has double significance. On the one hand, efficacy prediction is a prerequisite for achieving personalized precision medicine; on the other hand, it is also of great significance to understand the microscopic mechanism of SIIT in achieving diabetes remission by means of independent predictors of long-term remission after SIIT. At present, scholars at home and abroad have established a large number of clinical prediction models that can be used to predict the efficacy after SIIT, and mainly predict based on the following categories of clinical indicators: blood glucose indicators reflecting blood glucose control (such as time to achieve blood glucose target, 2-hour postprandial blood glucose, etc.), changes in insulin requirements, restoration of β-cell function and insulin sensitivity, and some serum biochemical markers. However, the existing SIIT remission indicators have limitations in clinical application. First of all, the clinical parameters for functional evaluation of islet or peripheral insulin sensitivity and DPP-4 can only be obtained after the end of the treatment cycle or after following up for a certain period, which may cause delays in clinical decision-making; secondly, dynamic indicators during treatment are easily interfered by external factors such as diet structure and exercise intensity; more importantly, the existing indicators cannot reveal the core molecular mechanism of β-cell function remodeling driven by SIIT, making the efficacy of the prediction model still only stay at the clinical prediction level and unable to indicate the molecular biological mechanism behind SIIT. This mechanism ambiguity restricts the accuracy of clinical decision-making, especially it is difficult to achieve individualized prognosis judgment in the face of complex cases.

[0005] MicroRNA (miRNA), as an endogenous non-coding RNA with a length of about 18 - 25 nucleotides, regulates the expression of most human genes at the post-transcriptional level. MicroRNA has a "single hit, multiple targets" effect, and a single microRNA can regulate hundreds of target genes, forming a complex multi-dimensional regulatory network. This characteristic enables it to play a key role in maintaining metabolic homeostasis and metabolic diseases such as diabetes involving multi-organ interactive lesions; in addition, microRNA is also very crucial in the local regulation of islet β-cell function, and its unique existence form in body fluids endows it with significant clinical biomarker characteristics. These unique molecular characteristics lay a solid theoretical foundation for the application of microRNA in clinical diagnosis and treatment monitoring. However, there is no study systematically exploring the role of microRNA in predicting the efficacy of SIIT, and this gap needs to be filled urgently. Summary of the Invention

[0006] In view of the above problems, the present invention aims to provide a biomarker for predicting the effect of intensive insulin therapy and its application. Through research, it is found that hsa-miR-502-3p is an independent predictor of long-term remission of SIIT, and a new combined prediction model for long-term remission of SIIT based on hsa-miR-502-3p and ISSI-2 is constructed; the regulatory relationship between hsa-miR-502-3p and SUR1 is clarified, revealing part of the molecular biological mechanism of the heterogeneity of SIIT treatment, providing a new perspective and basis for T2DM research and SIIT clinical practice.

[0007] To achieve the above object, the technical solutions adopted by the present invention are as follows:

[0008] On the one hand, the present invention provides a biomarker for predicting the effect of intensive insulin therapy, and the biomarker includes hsa-miR-502-3p in serum.

[0009] Furthermore, the biomarker further includes ISSI-2.

[0010] On the other hand, the present invention also provides a prediction model for predicting the effect of intensive insulin therapy, and the prediction model is constructed according to the biomarker as described above.

[0011] Furthermore, the prediction model includes a multi-factor logistic regression equation model and a nomogram model.

[0012] Furthermore, the nomogram model includes:

[0013] The first row is a score scale with a score range of 0 to 100;

[0014] The second row is the weight before intensive insulin therapy, with a range of 50 - 110 kg;

[0015] The third row is gender, including male and female;

[0016] The fourth row is age, with a range of 20 - 70 years old;

[0017] The fifth row is body mass index BMI, with a range of 20 - 34 kg / m 2 ;

[0018] The sixth row is glycated hemoglobin, with a range of 8% - 19%;

[0019] The seventh row is insulin sensitivity index - 2, with a range of 50 - 650;

[0020] The eighth row is hsa-miR-502-3P, with a range of -2.5 - 2.6;

[0021] The ninth row is the prediction score, ranging from 280 to 420;

[0022] The tenth row is the probability of maintaining remission after SIIT, ranging from 0% to 100%.

[0023] On the other hand, the present invention also includes the use of a reagent for detecting hsa-miR-502-3p in the preparation of a product for predicting the effect of intensive insulin therapy for diabetes.

[0024] Furthermore, the product includes a kit, a fluorescent probe, and a chip.

[0025] On the other hand, the present invention also includes the use of a small molecule inhibitor of the hsa-miR-502-3p / SUR1 pathway in the preparation of a drug for treating diabetes.

[0026] The beneficial effects of the present invention are as follows:

[0027] 1. Through research, the present invention found that baseline hsa-miR-502-3p is an independent predictor of long-term remission of SIIT, and successfully constructed a new combined prediction model for long-term remission of SIIT based on hsa-miR-502-3p and ISSI-2. The combined prediction model increased the ROC-AUC to 83.7%, an 8.8% increase compared to the single ISSI-2 model, further improving the model's recognition ability and achieving a better balance between prediction efficacy and clinical interpretability. At the same time, the clinical decision curve of the combined model indicates that at a 40% risk threshold, the combined model can avoid 69% of unnecessary subsequent SIIT interventions, significantly optimizing the allocation of medical resources. The online web prediction tool for long-term remission of SIIT developed based on the combined model realizes interactive visual risk assessment, providing technical support for precise intervention by medical institutions and clinicians.

[0028] 2. From the perspective of molecular mechanisms, this study also clarified the regulatory relationship between hsa-miR-502-3p and SUR1, and for the first time confirmed that hsa-miR-502-3p damages insulin secretion and cell proliferation and differentiation under high glucose stimulation of pancreatic islet β cells by targeting SUR1 / ABCC8; it reveals part of the molecular biological mechanism of SIIT treatment heterogeneity, providing a new perspective and basis for T2DM research and SIIT clinical practice.

[0029] 3. The research results of the present invention have multiple implications for clinical practice. First, baseline hsa-miR-502-3p detection can serve as a clinical tool for making precise intervention decisions after SIIT. For patients with a relatively high baseline hsa-miR-502-3p expression level, they should be advised to strengthen their own blood glucose detection and maintain closer follow-up observations to enhance the β-cell protective effect of SIIT. Second, T2DM patients with high baseline hsa-miR-502-3p expression show more significant β-cell function decline. Early identification for precise stratification of T2DM patients with different characteristics, and the development of small molecule inhibitors targeting the hsa-miR-502-3p / SUR1 pathway may become a new direction for diabetes reversal treatment and enhancing the efficacy of SIIT. Brief Description of the Drawings

[0030] Figure 1 It is a flow chart of the research process of the present invention.

[0031] Figure 2 It is the serum microRNA expression profile of patients before and after SIIT in the present invention.

[0032] Figure 3 It is the result of microRNA feature screening by LASSO regression, RF, and SVM-RFE in the present invention.

[0033] Figure 4 It is the Venn diagram of microRNA screened by three machine learning algorithms in the present invention.

[0034] Figure 5 It is the verification result of machine learning screening microRNA in an independent cohort in the present invention.

[0035] Figure 6 It is the change result of hsa-miR-502-3p in the serum of two groups of patients compared with themselves before receiving SIIT in the present invention.

[0036] Figure 7 It is the correlation analysis result between the expression level of hsa-miR-502-3p and glycemic homeostasis parameters in the present invention.

[0037] Figure 8 It is the evaluation result of the performance of three SIIT clinical prediction models constructed in the present invention.

[0038] Figure 9 It is the internal and external verification results of the combined model in the present invention.

[0039] Figure 10 It is the nomogram visualization model of the combined model in the present invention.

[0040] Figure 11This is the result of the enrichment analysis of the target genes of hsa-miR-502-3p in the present invention.

[0041] Figure 12 This is the result of verifying the overexpression efficiency of the in vitro model transfection in the present invention.

[0042] Figure 13 This is the result of the GSIS, CCK-8, and Edu experiments on the function of overexpressing mmu-miR-500-3p to damage MIN6 cells in the present invention.

[0043] Figure 14 This is the result of detecting the expression level of SUR1 after overexpressing hsa-miR-502-3 / mmu-miR-500-3p in HEK-293T cells and MIN6 cells in the present invention.

[0044] Figure 15 This is the result of the dual-luciferase reporter experiment to confirm that hsa-miR-502-3 targets the 3' UTR region of SUR1 in the present invention. Detailed implementation manners

[0045] In order to enable those of ordinary skill in the art to better understand the technical solution of the present invention, the technical solution of the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0046] 1 Research objects and methods

[0047] 1.1 Research objects and research methods

[0048] The subjects in this study were from two independent national multi-center randomized controlled trials (RCTs) led by the First Affiliated Hospital of Sun Yat-sen University. These two randomized controlled trials (NCT03194945 and NCT03972982) had similar study designs. NCT03194945 aimed to determine whether SIIT combined with metformin, linagliptin, or metformin plus linagliptin could maintain better long-term glycemic control compared to traditional SIIT treatment. NCT03972982 was to evaluate whether a simplified short-term insulin intensive treatment regimen was superior to traditional SIIT treatment in achieving long-term glycemic remission. In this study, NCT03194945 and NCT03972982 were used as two different sources of subjects, and the specific randomized controlled clinical trials were not the content of this study.

[0049] After admission, the patient first underwent measurements of routine body composition indices and biochemical indices, and assessments of blood sampling before and after the completion of a standard mixed meal tolerance test (MMTT). Subsequently, the patient received standardized continuous subcutaneous insulin infusion. During hospitalization, the insulin infusion was adjusted according to the capillary blood glucose results at 8 o'clock every day. The goal of blood glucose control was fasting plasma glucose (FPG) < 6.1 mmol / L and 2-hour postprandial blood glucose (2h-PG) < 8.0 mmol / L. After blood glucose reached the standard, the insulin infusion was continued for 14 days. After the end of SIIT, the patient's body composition and biochemical indices were measured and evaluated again. After the patient was discharged, regular follow-up was conducted to monitor the remission status of diabetes. If the patient had a recurrence of hyperglycemia, the patient was guided to receive hypoglycemic treatment according to the current guidelines. The long-term remission of diabetes after SIIT was defined as follows: within the first three months after SIIT intervention, FPG < 7 mmol / L; starting from the fourth month after SIIT, HbA1c < 6.5%. The remission of diabetes maintained for more than one year was defined as long-term remission. According to whether the definition of long-term remission was met, the patients were divided into a remission group (R) and a non-remission group (NR). In this study, signed informed consent was obtained from each patient and approval was obtained from the Research Ethics Committee of Sun Yat-sen University.

[0050] The research process of this study is as shown in the appendix Figure 1 below. The construction of the clinical prediction model for long-term remission of SIIT was divided into three stages. In stage 1, based on the serum microRNA expression profiles of 24 patients (discovery set, R:NR = 12:12) in the traditional SIIT treatment group in the NCT03194945 cohort, multiple machine learning algorithms were used for feature engineering optimization to screen for differentially expressed microRNAs related to long-term remission of SIIT. In stage 2, based on the sera of 97 patients (R:NR = 48:49) in the traditional SIIT treatment group from the NCT03972982 cohort, RT-qPCR was used to verify the differential microRNAs. In stage 3, the clinical data of 97 patients were split into a training set (70%) and a validation set (30%) according to a ratio of 7:3 for the construction of 3 clinical prediction models for the efficacy prediction of long-term remission of SIIT, and the performance of the models was evaluated from three dimensions: discrimination, calibration, and clinical utility of the SIIT model, and an interactive visualization of a web-based dynamic nomogram was performed for the model with the optimal performance. Based on the microRNAs closely related to the long-term remission of SIIT, through bioinformatics analysis, a list of target genes regulated by microRNAs and the results of enrichment analysis were obtained. The interaction relationship between microRNAs and mRNAs was verified through an in vitro cell model to explore the possible molecular mechanisms of SIIT heterogeneity.

[0051] 1.2 Experimental Materials and Methods

[0052] 1.2.1 Main Materials

[0053] 1.2.1.1 Clinical Serum Samples

[0054] Before and after SIIT, the routine biochemical indexes of patients were uniformly detected by a standardized laboratory in the testing center; after MMTT, blood samples of patients were collected on an empty stomach, at 30 minutes, 60 minutes, and 120 minutes to detect blood glucose and insulin levels. The insulin sensitivity of patients was evaluated by the homeostasis model assessment-insulin resistance (HOMA-IR) and Matsuda Index; the function of pancreatic islet β cells was evaluated by HOMA-β, insulin secretion-sensitivity index-2 (ISSI-2), and glucose disposition index (DI). The remaining blood samples were centrifuged, and the supernatants were aliquoted and stored. The freeze-thaw cycles of all blood samples did not exceed 3 times.

[0055] 1.2.1.2 Experimental Cells

[0056] The cell lines used in this study included human embryonic kidney HEK-293T cells and mouse pancreatic islet β cell line MIN6 cells, and the above cells have all been verified by Short Tandem Repeat (STR).

[0057] 1.2.2 Experimental Methods

[0058] 1.2.2.1 microRNA and Total RNA Extraction

[0059] Extract human serum microRNA according to the instructions of the microRNA extraction kit (QIAGEN); extract total RNA from cells using the Trizol method.

[0060] 1.2.2.2 Reverse Transcription Real-Time Fluorescent Quantitative PCR

[0061] (1) Reverse Transcription PCR

[0062] Perform microRNA reverse transcription according to the instructions of the microRNA reverse transcription kit (Aike Rui), and the microRNA reverse transcription system is shown in Table 1 below.

[0063] Table 1 microRNA Reverse Transcription System

[0064]

[0065] Cell RNA reverse transcription was performed according to the instructions of the RNA reverse transcription kit (Promega). The reverse transcription system of total cellular RNA is shown in Table 2 below.

[0066] Table 2 Reverse transcription system of total cellular RNA

[0067]

[0068] The above components were mixed well on ice and centrifuged instantaneously, and then the reverse transcription reaction was carried out. After the reverse transcription was completed, the products were collected for subsequent real-time fluorescence quantitative PCR reaction.

[0069] (2) Real-time fluorescence quantitative PCR

[0070] The primers were designed using the NCBI Primer Blast online tool. The sequences of the real-time fluorescence PCR primers used in this study are shown in Table 3 below. According to the instructions of the kit (Aike Rui), the reaction system was prepared on an ice box. The cDNA real-time fluorescence quantitative PCR reaction system is shown in Table 4 below.

[0071] Table 3 Primer sequences used in real-time fluorescence quantitative PCR

[0072]

[0073] Table 4 cDNA real-time fluorescence quantitative PCR reaction system

[0074]

[0075] After the cDNA real-time fluorescence quantitative PCR reaction system was added to the 96-well plate, it was covered with a film, sealed, and centrifuged instantaneously. After centrifugation, the system was placed into the real-time fluorescence quantitative PCR instrument. The preset reaction program was as follows: pre-denaturation (Step 1: 95°C, 30 seconds); PCR reaction (Step 2: 95°C, 5 seconds, 60°C, 30 seconds, a total of 40 cycles); melting curve analysis (Step 3: 95°C, 15 seconds, 60°C, 60 seconds, 95°C, 15 seconds).

[0076] (3) Result analysis

[0077] The relative expression level of the target gene was calculated according to the relative quantification method. When analyzing the microRNA expression level in human serum samples, cel-miR-39-3p was used as the internal reference, and the patients in the NR group were used as the control group; when analyzing the RNA expression level in cells, BETA-ACTIN was used as the internal reference, and the cells in the control group were used as the control group.

[0078] 1.2.2.3 Cell culture

[0079] (1) Cell resuscitation

[0080] Take out the cell cryopreservation tube from liquid nitrogen, quickly thaw it in a 37°C water bath with shaking until at least a small amount of ice crystals melt. Centrifuge the cell cryopreservation tube at 950 rpm for 3 minutes. After centrifugation, take out the cryopreservation tube and transfer it to a biosafety cabinet. Discard the supernatant, resuspend the cells with pre-warmed complete medium, inoculate them into a culture dish, and culture at 37°C and 5% CO2. Change the medium after 24 hours.

[0081] (2) Cell culture and passage

[0082] MIN6 cells are cultured in high-glucose DMEM + 15% FBS + 50 μmol / L β-mercaptoethanol + 1% streptomycin-penicillin system, and the medium is changed every 2 days; 293T cells are cultured in high-glucose DMEM + 10% FBS + 1% streptomycin-penicillin system, and observed daily.

[0083] During passage, digest with an appropriate amount of 0.25% trypsin for 2 minutes (37°C), add 2 volumes of medium to terminate digestion, and passage according to a ratio of 1:3 - 1:5 based on the cell number after centrifugation; MIN6 cells adhere loosely, avoid over-pipetting, and maintain the cell cluster structure; 293T cells adhere tightly, with a passage interval of 48 - 72 hours to avoid over-confluence.

[0084] (3) Cell cryopreservation steps

[0085] After digesting and centrifuging the cells, count them using a cell counting chamber. After adjusting the cell suspension to an appropriate density, aliquot it into internal-rotation cell cryopreservation tubes, use a cryopreservation box to store at -80°C overnight, and then transfer it to liquid nitrogen for long-term storage.

[0086] 1.2.2.4 Transient transfection of cells

[0087] MicroRNA overexpression mimics and negative control oligonucleotides (Negative Control, NC) are purchased from Ribobio. The transient transfection experiment of cells is carried out using the ribOFECT CP transfection kit (Ribobio). The groups are divided into a control group (Control), an overexpression group (mimic), and a negative control group (NC). The specific transient transfection process is not described in detail in this invention.

[0088] 1.2.2.5 Cell protein extraction and protein quantification

[0089] The operations of cell protein extraction and protein quantification include the steps: (1) cell lysis treatment; (2) separation of protein supernatant; (3) protein concentration determination; (4) protein denaturation and storage. The specific operation process of each step is prior art and is not described in detail in this invention.

[0090] 1.2.2.6 Western blot experiment

[0091] (1) Preparation of protein samples

[0092] Take out the protein sample to be tested, thaw it at room temperature, and calculate the sample loading amount per well (50 μg) according to the protein concentration obtained by BCA quantification;

[0093] (2) Electrophoresis separation and membrane transfer

[0094] Perform SDS-PAGE vertical electrophoresis using a 12.5% separating gel and a 5% stacking gel system. Set the initial voltage at 80 V during the stacking stage, and adjust it to 120 V after the protein and the maker enter the separating gel. After electrophoresis, transfer the protein to a 0.22-μm PVDF membrane at a constant current of 300 mA through a wet transfer system. The PVDF membrane is pre-activated by soaking it in methanol for 1 minute;

[0095] (3) Construction of the immune reaction system

[0096] After membrane transfer, block it with 5% non-fat milk-TBST blocking solution at room temperature for 1 hour. Subsequently, incubate the protein with a specifically prepared primary antibody working solution (antibody stock solution: antibody diluent = 1:1000), and incubate it overnight on a shaker in a 4°C refrigerator. After overnight incubation, wash away the non-specific binding background with TBST, add the corresponding species-specific HRP-labeled secondary antibody (antibody stock solution: antibody diluent = 1:5000) prepared, and incubate it at room temperature for 2 hours;

[0097] (4) Signal detection and analysis

[0098] Cover the surface of the membrane evenly with ECL chemiluminescent substrate, and collect the signal using an exposure instrument. Standardize the expression level of the target protein using BETA-ACTIN as an internal reference, and perform gray value analysis based on the Image J software.

[0099] 1.2.2.7 Static insulin secretion stimulation experiment (Glucose-stimulated insulin secretion, GSIS)

[0100] (1) After MIN6 cells are balanced with KRB buffer (containing 2 mmol / L glucose and 1 g / L BSA) for 1 hour, stimulate insulin secretion with KRB buffer containing 2 mmol / L and 20 mmol / L glucose respectively;

[0101] (2) After incubation at 37°C for 1 hour, collect the supernatant and remove cell debris by low-temperature centrifugation at 4°C; Extract and quantify the total protein using RIPA lysis buffer and BCA quantification method;

[0102] (3) Use a mouse ultrasensitive ELISA kit (Mercodia) to measure the insulin concentration, calculate the insulin concentration through the standard curve equation, and standardize it with the total protein concentration.

[0103] 1.2.2.8 CCK8 Assay

[0104] (1) Observe the cell growth status, inoculate the cells to be treated with good status into a 96-well plate, set up a blank control and transfection concentration gradient treatment groups, and wait for the cells to adhere to the plate;

[0105] (2) After 48 hours of transfection, add the prepared CCK-8 working solution (culture medium: CCK-8 mother liquor = 10:1) to each well, and incubate at 37 °C in a cell incubator for 2 hours;

[0106] (3) Measure the absorbance at the corresponding wavelength with an enzyme-labeled instrument, and subtract the background value of the blank control group during data processing.

[0107] 1.2.2.9 Edu Assay

[0108] (1) Discard the old culture medium of the treated cells, and co-incubate with the complete culture medium containing 15 μmol / L EdU for 2 hours;

[0109] (2) Fix the cells with 4% paraformaldehyde for 20 minutes, and permeabilize the cells with 0.5% Triton X-100 for 20 minutes;

[0110] (3) Add 200 μL of Click-iT reaction solution to the system, keep it in the dark at room temperature for 30 minutes, observe with a fluorescence microscope after Hoechst counterstaining the cell nuclei, and quantitatively analyze the positive rate with ImageJ.

[0111] 1.2.2.10 Dual-Luciferase Reporter Assay

[0112] (1) Clone the wild-type and mutant 3' UTR sequences of the target gene SUR1 into the pmirGLO vector, and co-transfect HEK-293T cells with the microRNA mimic and the negative control group (NC);

[0113] (2) After 48 hours of transfection, take out the cells to be tested and equilibrate them to room temperature at 25 °C, and let them stand for about 30 minutes;

[0114] (3) Measure the firefly luciferase activity: Add 100 μL of Duo-Lite Luciferase reagent equal in volume to the culture system, place it for 10 minutes, and measure the luminescence intensity of the firefly luciferase with a multi-functional chemiluminescence enzyme-labeled instrument;

[0115] (4) Measure the Renilla luciferase activity: Add 100 μL of Duo-Lite Stop&Lite reagent equal in volume to the culture system; place it for 10 minutes, and detect the luminescence of the Renilla luciferase with a multi-functional chemiluminescence enzyme-labeled instrument; The final reported activity is expressed as the ratio of firefly luciferase to Renilla luciferase.

[0116] 1.2.2.11 Serum small RNA sequencing

[0117] (1) RNA extraction and quality assessment: Total RNA was extracted from serum samples using TRIzol reagent, and the RNA integrity was detected by an Agilent 2100 bioanalyzer. The RNA integrity number (RIN) should be greater than 7;

[0118] (2) Construction of small RNA library: Adapter ligation was performed at the 3' and 5' ends in sequence. Reverse transcription and PCR amplification were carried out on the small RNA molecules that had completed double-end ligation. The target fragments (140 - 160 bp) were separated by agarose gel electrophoresis and purified by gel cutting;

[0119] (3) High-throughput paired-end sequencing: Library construction and high-throughput sequencing were completed based on the Illumina HiSeq Xten platform.

[0120] 1.2.2.12 Bioinformatics data analysis

[0121] (1) Preprocessing and quality control of raw data: After the raw sequencing data was downloaded, FastQC (v0.12.1) was used for quality assessment and preprocessing,

[0122] and the screening criteria were set as the sequence length not less than 18 nucleotides (nt) to ensure obtaining high-quality clean reads;

[0123] (2) Comparison and annotation of sequences: Feature analysis of small RNA molecules was carried out, and systematic identification and functional annotation of the tag sequences of microRNA were focused on;

[0124] (3) Expression analysis: miRDeep2 was used for microRNA quantification to obtain the microRNA expression matrix of all samples; by setting the differential threshold of FDR < 0.05 and |log2FC| > 1.5, a list of differentially expressed microRNAs was obtained; the FactoMineR package was used for principal component analysis (PCA), and the heat map was used to show the differential microRNA expression levels among different samples;

[0125] (4) Functional enrichment analysis: Miranda (v3.3a), TargetScan (v7.0) and RNAhybrid (v2.1) were integrated to predict potential target mRNAs of microRNA. Gene Ontology (GO) and Kyoto Encyclopedia of Genes and Genomes pathway enrichment analysis (KEGG) were used to obtain the biological pathways and processes regulated by target genes;

[0126] (5) The above analysis was performed based on the R 4.2.1 software, and the Benjamini-Hochberg method was used for multiple corrections.

[0127] 1.3 Construction and statistical methods of the clinical prediction model

[0128] 1.3.1 Screening of microRNA biomarkers and clinical indicators

[0129] 1.3.1.1 Screening of microRNA related to long-term remission of SIIT

[0130] Since there are numerous microRNAs with changed expressions in the sera of patients before and after SIIT, in this study, three machine learning algorithms were used to optimize the feature engineering of high-dimensional omics data (discovery set, N = 24) to screen for microRNAs closely related to long-term remission after SIIT, including Least Absolute Shrinkage and Selection Operator regression (LASSO), Random Forest (RF), and Support Vector Machines-Recursive Feature Elimination (SVM-RFE). Through the above steps of joint screening, the common microRNAs identified by all methods were established as the key predictors of long-term remission of SIIT.

[0131] 1.3.1.2 Screening of clinical indicators related to long-term remission of SIIT

[0132] According to the clinical data of patients before and after SIIT, candidate clinical variables were initially screened by univariate analysis: for continuous variables, Student’s t-test or Wilcoxon rank sum test was used for inter-group difference analysis; for categorical variables, Pearson chi-square test or Fisher's exact probability method was used for statistical evaluation; a univariate logistic regression equation for long-term remission of SIIT was initially constructed, and the predictors with statistical significance (P < 0.1) were included in the alternative variables. Subsequently, the two-way stepwise regression method was used to perform a secondary filtration of the alternative variables. This algorithm optimizes the prediction efficacy while controlling the model complexity through an iterative process of forward selection and backward elimination, and finally obtains the clinical indicators related to long-term remission of SIIT.

[0133] 1.3.2 Construction of the clinical prediction model

[0134] Based on the microRNAs and clinical indicators screened in the early stage, this study finally constructed the following three clinical prediction models for predicting the long-term remission of SIIT through multivariate logistic regression:

[0135] Model 1 (Clinical Indicator Model): It includes the screened clinical features and correction variables;

[0136] Model 2 (microRNA Model): It includes the screened microRNAs;

[0137] Model 3 (Combined Model): It includes the screened clinical features, microRNAs and correction variables.

[0138] 1.3.3 Evaluation of the Performance of Clinical Prediction Models and Development of Dynamic Nomograms

[0139] To comprehensively evaluate the performance of the constructed SIIT prediction models, this study evaluated the models from three aspects: discrimination, calibration, and clinical utility. The model with the best performance was visualized with a static nomogram and a web-based dynamic nomogram.

[0140] Discrimination was evaluated by calculating and comparing the area under the receiver operating characteristic curve (ROC-AUC), which can effectively reflect the model's ability to identify long-term remission of SIIT. Calibration was visualized using a locally weighted scatterplot smoothing (LOESS) calibration curve, which reflects the matching degree between the predicted probability and the actual observed value through non-parametric regression techniques. In terms of clinical utility assessment, dual validation of decision curve analysis (DCA) and clinical impact curve (CIC) was introduced. Decision curve analysis quantifies the clinical net benefit of the SIIT model through threshold probability analysis, and the clinical impact curve dynamically demonstrates the patient classification effect at different risk thresholds. At the same time, diagnostic indicators such as model sensitivity, specificity, negative predictive value (NPV), and positive predictive value (PPV) were calculated to improve the evaluation system. During the model validation stage, internal validation was performed based on the training set data, and external generalization ability testing was performed based on the validation set data. To compensate for potential model overfitting and evaluation bias caused by a small sample size, this study used Bootstrap resampling for correction, that is, 1000 Bootstrap samples were generated for each multiply imputed dataset, and a validation framework was constructed through repeated sampling. During the model visualization stage, based on the Shiny framework (version 0.13.2.26), an interactive web-based dynamic nomogram platform for predicting long-term remission of SIIT was developed, and individualized risk prediction was achieved through a visual interface, thus improving the accessibility and practicality of the clinical decision support system.

[0141] 1.3.4 Statistical methods

[0142] In this study, standard statistical methods were used to analyze and process the data. Continuous variables that conform to the normal distribution were presented in the form of mean ± standard deviation (Mean±SD), and one-way analysis of variance (One-way ANOVA) or Student's t-test was used for comparison between groups; data with non-normal distribution were expressed as median (interquartile range), and Kruskal-Wallis H test or Mann-Whitney U test was selected for difference analysis. The chi-square test was used for comparison of categorical variables. Pearson correlation or Spearman correlation analysis was used for correlation analysis between variables according to whether the data obeyed the normal distribution. All change values (delta values) in this study were defined as the difference between the measured values after SIIT and those before SIIT. In terms of experimental design, in vitro experiments strictly followed the principle of repeatability, and each experimental condition was independently repeated at least three times. The specific number of repetitions was clearly marked in the corresponding figure legends. The above analysis was mainly based on R (version 4.2.1), and GraphPad Prism 9.0 was used to complete data visualization and auxiliary analysis. The statistical significance threshold was set at P≤0.05, and all tests were two-sided tests.

[0143] 2 Research Results

[0144] 2.1 Queue Characteristics

[0145] The clinical data of the patients in the SIIT with pump treatment group in the two RCT queues included in this study before and after SIIT are statistically shown in Tables 1 and 2 below. Among them, Table 1 shows the changes in blood glucose and islet function indexes before and after SIIT in the remission group and non-remission group in the discovery set, and Table 2 shows the changes in blood glucose and islet function indexes before and after SIIT in the remission group and non-remission group in the training set and validation set.

[0146] In Table 1 and Table 2, the meanings of different indicators are as follows: body mass index (BMI), glycated hemoglobin A1c (HbA1c), fasting plasma glucose (FPG), 2-hour post-prandial plasma glucose (2h-PG), homeostasis model assessment (HOMA), HOMA-beta, Disposition Index, HOMA-IR, Insulin Secretion-Sensitivity Index-2 (ISSI-2), R (remission group), NR (non-remission group), delta value = after SIIT - before SIIT.

[0147] As can be seen from Table 1, for the 24 patients (discovery set, NCT03194945) used for small-RNA sequencing, the baseline characteristics between the two groups were balanced and comparable. After SIIT treatment, FPG decreased significantly from the baseline value (10.89 ± 2.48 mmol / L) to 6.17 ± 1.38 mmol / L; 2h-PG decreased from 19.03 ± 3.20 mmol / L to 14.17 ± 2.98 mmol / L (P < 0.05). At the same time, a series of indicators reflecting β-cell function showed significant improvement, including the homeostasis model β-cell function index (HOMA-β), glucose disposition index (DI), and insulin secretion sensitivity index 2 (ISSI-2).

[0148] As can be seen from Table 2, for the 97 patients (training set and validation set, NCT03972982) used for RT-qPCR verification, except for the difference in baseline weight between groups (P = 0.041), the remaining parameters were balanced between groups, and this covariate was corrected in the multivariate regression model. After SIIT, both FPG and 2h-PG in the two groups of patients decreased significantly. At the same time, the remission group showed better ISSI-2 and disposition index (DI) after treatment, and their change values (ΔISSI-2 and ΔDI) were also significantly higher than those of the non-remission group.

[0149] Table 1 Changes in blood glucose and islet function indicators before and after SIIT treatment in the remission group and non-remission group of the discovery set

[0150]

[0151] Table 2 Changes in blood glucose and islet function indexes before and after SIIT in the remission group and non-remission group of the training set and validation set

[0152]

[0153] Continued Table 2 Changes in blood glucose and islet function indexes before and after SIIT in the remission group and non-remission group of the training set and validation set

[0154]

[0155] 2.2 Changes in serum microRNA expression in patients before and after SIIT

[0156] Analysis of small-RNA omics data showed that there were 46 microRNAs with baseline differential expression in the remission group compared to the non-remission group, among which 35 were significantly upregulated and 11 microRNAs were significantly downregulated. Among them, the basally upregulated microRNAs included: hsa-miR-532-5p, hsa-miR-660-5p, hsa-miR-6764-5p, hsa-miR-6767-5p, hsa-miR-199a-5p, hsa-miR-487b-3p, hsa-miR-130b-5p, hsa-miR-3157-5p, hsa-miR-92a-1-5p, hsa-let-7i-3p, hsa-miR-6842-5p, hsa-miR-454-5p, hsa-miR-1306-3p, hsa-miR-26b-3p, hsa-miR-6747-3p, hsa-miR-542-5p, hsa-miR-190a-5p, hsa-miR-1273h-3p, hsa-miR-141-3p, hsa-miR-378a-5p, hsa-miR-7-1-3p, hsa-miR-301a-3p, hsa-miR-374b-5p, hsa-miR-29c-5p, hsa-miR-551a, hsa-miR-199b-5p, hsa-miR-3074-5p, hsa-miR-505-3p, hsa-miR-185-3p, hsa-miR-21-3p, hsa-miR-379-5p, hsa-miR-19a-3p, hsa-miR-22-5p, hsa-miR-365a-3p, hsa-miR-365b-3p; the basally downregulated microRNAs included: hsa-miR-363-3p, hsa-miR-4685-3p, hsa-miR-296-5p, hsa-miR-636, hsa-miR-6852-5p, hsa-miR-195-5p, hsa-miR-200a-3p, hsa-miR-502-3p, hsa-miR-3613-5p, hsa-miR-196a-5p, hsa-miR-6803-3p.

[0157] After SIIT intervention, the number of differentially expressed microRNAs between groups decreased to 42, among which 31 were significantly up-regulated and 11 were significantly down-regulated in the remission group. Among them, the up-regulated microRNAs included: hsa-miR-3143, hsa-miR-11401, hsa-miR-32-5p, hsa-miR-501-5p, hsa-miR-6852-5p, hsa-miR-542-5p, hsa-miR-363-3p, hsa-miR-3613-3p, hsa-miR-181c-5p, hsa-miR-551a, hsa-miR-301a-3p, hsa-miR-7-1-3p, hsa-miR-548j-5p, hsa-miR-4435, hsa-miR-92a-1-5p, hsa-miR-128-1-5p, hsa-miR-4286, hsa-miR-30c-5p, hsa-miR-221-5p, hsa-miR-4714-3p, hsa-miR-1260b, hsa-miR-6862-5p, hsa-miR-4685-3p, hsa-miR-654-3p, hsa-miR-1273h-3p, hsa-miR-204-3p, hsa-miR-30b-5p, hsa-miR-505-3p, hsa-miR-29c-5p, hsa-miR-222-3p, hsa-miR-190a-5p; the down-regulated microRNAs included: hsa-miR-374a-5p, hsa-miR-874-3p, hsa-miR-369-5p, hsa-miR-141-3p, hsa-miR-374b-5p, hsa-miR-3158-3p, hsa-miR-636, hsa-miR-323a-3p, hsa-miR-9-5p, hsa-miR-6891-5p, hsa-miR-204-5p.

[0158] Principal component analysis (PCA) and heat map analysis were performed on the serum microRNAs of patients before and after SIIT, and the expression profiles of serum microRNAs of patients before and after SIIT were obtained as attached Figure 2 shown. Among them, A and B are the results of principal component analysis of serum microRNAs of patients before and after SIIT treatment, respectively; C and D are the heat map distributions of serum microRNAs of patients before and after SIIT treatment, respectively. It can be seen from the attached Figure 2 that the samples of the remission group and the non-remission group before treatment were spatially separated in the dimensions of PC1 (explaining 86.57% of the variance) and PC2 (explaining 5.81% of the variance), indicating that there was a certain heterogeneity in the microRNA expression profiles of the two groups at baseline (as attachedFigure 2 In A). The PCA distributions of the two groups of samples after SIIT treatment showed a convergent trend, and the differences between groups decreased (as shown in the appendix Figure 2 In B). The heatmap distribution of differentially expressed microRNAs is shown in the appendix Figure 2 in C and D.

[0159] 2.3 Screening of differentially expressed microRNAs related to long-term remission of SIIT by machine learning

[0160] Feature engineering filtering of microRNAs was performed using three machine learning algorithms, LASSO regression, RF, and SVM-RFE, to screen for differentially expressed microRNAs most closely related to long-term remission of SIIT. The results are shown in the appendix Figure 3 as follows. A shows the number of subsets corresponding to λ = minimum and λ = 1se, which are 9 and 1 microRNA, respectively; B shows the convergence path diagram of LASSO regression coefficients; C shows the top 15 key microRNAs evaluated by the random forest MDA method; D shows the top 15 key microRNAs evaluated by the random forest MDG method; E shows the subset of 9 microRNAs when the support vector machine model reaches the optimal resolution performance; F shows the change in the resolution accuracy of the support vector machine model with feature iteration.

[0161] In the LASSO screening stage, all differentially expressed microRNAs were incorporated into the LASSO regularization penalty formula. Ten-fold cross-validation was used to reduce errors, and the feature subsets corresponding to λ = 1se and λ = minimum were selected. A total of 10 microRNAs related to long-term remission of SIIT were obtained, as shown in Figure 3 A and B in the appendix. In the RF feature screening, the mean decrease in accuracy (MDA, as shown in Figure 3 C in the appendix) and the mean decrease in Gini index (MDG, as shown in Figure 3 D in the appendix) were used to evaluate the importance of microRNA features. The results showed that the top 15 microRNAs selected by the two methods had significant consistency (11 overlaps), indicating high reliability of feature selection. In the SVM-RFE screening stage, when the microRNAs were iterated to include the 9 microRNA features shown in Figure 3 E in the appendix, the support vector machine model achieved the optimal performance for classifying SIIT patients, as shown in Figure 3 F in the appendix.

[0162] Combining the results of the above three machine learning algorithms, 5 common microRNAs were screened out, as shown in the appendix Figure 4As shown, the five common microRNAs, including hsa-miR-363-3p, hsa-miR-502-3p, hsa-miR-532-5p, hsa-miR-6891-5p, and hsa-miR-204-3p, entered the subsequent RT-qPCR verification stage.

[0163] 2.4 Independent cohort verification of differential microRNAs

[0164] 2.4.1 RT-qPCR verification of differential expression of microRNAs before and after SIIT

[0165] After the above microRNA screening process, five microRNAs (hsa-miR-363-3p, hsa-miR-502-3p, hsa-miR-532-5p, hsa-miR-6891-5p, and hsa-miR-204-3p) that may be related to the long-term remission of SIIT were finally obtained. Technical verification was carried out using RT-qPCR in an independent cohort (training set and validation set, N = 97), and the results are as follows Figure 5 As shown, among them, A: RT-qPCR showed that the abundance of hsa-miR-502-3p in the serum of patients in the remission group was lower than that in the non-remission group before receiving SIIT (N = 97); B: RT-qPCR showed that there was no significant difference in the abundance of serum hsa-miR-502-3p between groups after receiving SIIT (N = 97); C: RT-qPCR verification results of hsa-miR-363-3p, hsa-miR-532-5p, hsa-miR-6891-5p, and hsa-miR-204-3p in the independent cohort, with no significant difference between groups (N = 97). When calculating Log2FC, the CT value of Cel-miR-39-3p was used as the internal reference, and the non-remission group was used as the control. *P < 0.05; **P < 0.01; ***P < 0.001; ns, non-significant.

[0166] Appendix Figure 5 showed that compared with the non-remission group, the expression level of hsa-miR-502-3p in the remission group was significantly downregulated before SIIT (P = 0.0098) (as shown in A of the appendix Figure 5 ), and there was no statistical difference in the expression level of hsa-miR-502-3p after SIIT (P = 0.0503) (as shown in B of the appendix Figure 5 ), which was consistent with the sequencing results. The other four microRNAs did not show differences between groups before and after SIIT treatment (as shown in C of the appendix Figure 5 ).

[0167] 2.4.2 Calculate the changing trend of hsa-miR-502-3p within the group

[0168] Calculate the changes in serum hsa-miR-502-3p in the two groups of patients compared with themselves before receiving SIIT. The results are shown in the appendix Figure 6 As shown, when calculating Log2FC, the CT value of Cel-miR-39-3p was used as the internal reference, and the abundance of hsa-miR-502-3p in the patient's own serum before SIIT was used as the control. *P<0.05; **P<0.01; ***P<0.001. It can be seen from the appendix Figure 6 that when using the abundance of hsa-miR-502-3p in the patient's own serum before receiving SIIT as the control, the levels of hsa-miR-502-3p in both groups of subjects showed a downward trend after SIIT treatment, but this change was only statistically significant in the non-remission group (P = 0.028), while there was no significant change in the remission group (P = 0.056).

[0169] 2.5 Correlation between hsa-miR-502-3p and glycemic homeostasis parameters

[0170] This study further explored the association between the expression level of hsa-miR-502-3p and β-cell function indices (HOMA-β, ISSI-2, and disposition index DI), insulin resistance indices (HOMA-IR and Matsuda index), and blood glucose indices (FPG and 2h-PG). The results are shown in the appendix Figure 7 As shown, where A is the correlation analysis between baseline hsa-miR-502-3p and glucose after SIIT treatment; B is the correlation analysis between baseline hsa-miR-502-3p and post-ISSI-2 after SIIT treatment; C is the correlation analysis between baseline hsa-miR-502-3p and the edge of post-DI after SIIT treatment; D is the correlation analysis between hsa-miR-502-3p after SIIT treatment and MMTT-Ins60 after SIIT treatment.

[0171] It can be seen from the appendix Figure 7 that the baseline expression level of hsa-miR-502-3p was positively correlated with the area under the glucose curve (AUCglu) after SIIT (r = 0.21, P = 0.041), as shown in A of the appendix Figure 7 and negatively correlated with post-ISSI-2 after intervention (r = -0.21, P = 0.045), as shown in B of the appendix Figure 7 and marginally negatively correlated with post-DI after intervention (r = -0.19, P = 0.058), as shown in C of the appendix Figure 7As shown in Figure C; the above results suggest that the lower the abundance of hsa-miR-502-3p before the patient receives SIIT treatment, the better the function of pancreatic islet β cells may recover after treatment. The abundance of hsa-miR-502-3p after SIIT intervention was negatively correlated with the insulin level at 60 minutes of MMTT (post-Ins60) (r=-0.20, P=0.047), as shown in Appendix Figure 7 Figure D.

[0172] In addition, according to the results of the correlation analysis between miR-502-3p and clinical data (shown in Table 3 below), no significant correlation was found between the level of hsa-miR-502-3p and insulin resistance-related indicators.

[0173] Table 3 Results of the correlation analysis between miR-502-3p and clinical data

[0174]

[0175] Continued Table 3 Results of the correlation analysis between miR-502-3p and clinical data

[0176]

[0177] Continued Table 3 Results of the correlation analysis between miR-502-3p and clinical data

[0178]

[0179] Note: Pearson correlation or Spearman correlation was used to evaluate the correlation between variables according to whether the variables were normally distributed.

[0180] 2.6 Construction of a clinical prediction model for SIIT and evaluation of model performance

[0181] 2.6.1 Construction of three clinical prediction models for predicting long-term remission of SIIT

[0182] The three SIIT long-term remission prediction models constructed in this study based on the multiple factor logistic regression equation are as follows:

[0183] Model 1 (clinical index model): includes the screened clinical features ISSI-2 and correction variables, as shown in Table 4 below; quantitatively expressed as Remission = 1.01*post-ISSI-2 + 1.083*pre-Weight + 1.221*pre-HbA1c + 0.852*pre-BMI + 1.020*Age + 2.270*Gender - 3.62.

[0184] Model 2 (miR model): It includes the screened hsa-miR-502-3p; quantitatively expressed as Remission = -1.07 * hsa-miR-502-3p - 0.068.

[0185] Model 3 (combined model): It includes the screened ISSI-2, hsa-miR-502-3p and correction variables, as shown in Table 5 below; quantitatively expressed as Remission = 0.257 * hsa-miR-502-3p + 1.01 * post-ISSI-2 + 1.006 * pre-Weight + 1.733 * pre-HbA1c + 1.174 * pre-BMI + 1.030 * Age + 1.600 * Gender - 3.26.

[0186] Table 4 Traditional clinical index SIIT model (Model 1)

[0187]

[0188] Table 5 Combined model of ISSI-2 and hsa-miR-502-3p (Model 3)

[0189]

[0190] 2.6.2 Evaluation of the performance of clinical prediction models

[0191] To comprehensively evaluate the accuracy of the constructed prediction model for predicting long-term remission after SIIT, this study plotted the Receiver Operating Characteristic curve (ROC), Calibration curve (CC), Clinical Decision Curve (DCA) and Clinical Impact Curve (CIC), as shown in the appendix Figure 8 Shown. Among them, A and B are respectively the Receiver Operating Characteristic curve and Calibration curve of the 3 clinical prediction models for long-term remission of SIIT; C is the model evaluation index, and D and E are respectively the Clinical Decision Curves of the 3 clinical prediction models for long-term remission of SIIT in the training set (D) and validation set (E); F is the Clinical Decision Curve of the combined model.

[0192] As can be seen from A in the appendix Figure 8 The ROC-AUC of the combined model reached 0.837 (95% CI: 0.732 - 0.917), which was significantly better than that of Model 1 (0.769, 95% CI: 0.649 - 0.865) and Model 2 (0.743, 95% CI: 0.624 - 0.856). To compare the goodness of fit of the models, this study also performed a likelihood ratio test on Model 1, Model 2 and Model 3. The results showed that the goodness of fit of Model 3 was significantly better than that of Model 1 (χ 2 = 13.955, P < 0.001). In addition, the difference between Model 3 and Model 2 was also significant (χ2 = 17.128, P = 0.009), indicating that the combination of ISSI-2 and hsa-miR-502-3p significantly improved the ability of the model to predict long-term remission of SIIT.

[0193] In terms of calibration (see Figure 8 B in the appendix), the Brier score of the combined model was 0.160 (95% CI 0.115 - 0.205), showing better consistency than Model 1 (Brier Score = 0.196, 95% CI 0.153 - 0.239) and Model 2 (Brier Score = 0.206, 95% CI 0.166 - 0.246). At the same time, under most risk thresholds, the combined Model 3 performed well in terms of specificity, sensitivity, negative predictive value (NPV), and positive predictive value (PPV) (see Figure 8 C in the appendix), suggesting good clinical applicability. In terms of clinical utility, the results of decision curve analysis showed that the net benefit of the combined model was significantly better than that of other models in most risk probability ranges, whether in the training set (see Figure 8 D in the appendix) or the validation set (see Figure 8 E in the appendix). Clinical impact curve analysis (see Figure 8 F in the appendix) further confirmed that the prediction results of the combined model had good consistency with the actual clinical outcomes.

[0194] In addition, this study further evaluated the performance of the combined model (Model 3) in internal and external validations, and the results are shown in the appendix Figure 9 as follows. In the appendix Figure 9 , A is the average discrimination of the combined model in the training set by Bootstrap resampling, and B is the average discrimination of the combined model in the validation set by Bootstrap resampling. The results of Bootstrap resampling internal validation showed that the mean ROC-AUC of the combined model was 0.845 (see Figure 9 A in the appendix). Using the same method in the external validation stage, the results showed that the mean ROC-AUC of the combined model was 0.893 (see Figure 9 B in the appendix).

[0195] 2.6.3 Visualization of the web-based dynamic nomogram of the combined model

[0196] This study constructed a visual nomogram for the combined Model 3, as shown in the appendix Figure 10 , and developed an online prediction tool based on the R-Shiny platform for the convenience of clinicians.

[0197] 2.7 hsa-miR-502-3p targets SUR1 to affect islet beta cell function

[0198] 2.7.1 Prediction and enrichment analysis results of hsa-miR-502-3p target genes

[0199] Integrate Miranda, TargetScan and RNAhybrid to predict the potential target genes of hsa-miR-502-3p, and 1845 potential target genes are obtained. Enrichment analysis is performed on the obtained potential target genes, and the results are as follows Figure 11 shown. Among them, A is the result of KEGG enrichment analysis, and B is the result of GO enrichment analysis. The results of KEGG pathway enrichment analysis show that the target genes are significantly aggregated in the metabolic-related pathways, and the enrichment degrees of the two signal pathways of type 2 diabetes mellitus and insulin secretion are the most prominent. The results of GO enrichment analysis indicate that the target genes are significantly enriched in biological processes such as potassium ion transport and ATPase-dependent transmembrane transport complex, and are significantly enriched in sulfonylurea receptor activity at the molecular function level. It is speculated that hsa-miR-502-3p may affect the intercellular exchange of potassium ions and calcium ions by targeting sulfonylurea receptor 1 (SUR1) on the islet β cell membrane, thereby affecting insulin secretion.

[0200] 2.7.2 hsa-miR-502-3p affects insulin secretion and beta cell proliferation and differentiation

[0201] To further verify whether hsa-miR-502-3p can affect insulin secretion, this study further carried out transfection experiments in human HEK-293T cells and mouse insulinoma MIN6 cell lines. Sequence alignment shows that human hsa-miR-502-3p has the exactly same nucleotide sequence (AAUGCACCUGGGCAAGGGUUCA) as mouse mmu-miR-500-3p. The results of verifying the overexpression efficiency of transfection are as follows Figure 12 shown. Among them, A is the efficiency of overexpression verified by RT-qPCR after transfection of the indicated concentration of microRNA mimic and negative control into HEK-293T cells, and B is the efficiency of overexpression verified by RT-qPCR after transfection of the indicated concentration of microRNA mimic and negative control into MIN6 cells. It can be seen from the following Figure 12 that the overexpression of hsa-miR-502-3p is successful in the in vitro model (N = 9).

[0202] The experimental results of the functional damage of overexpressing mmu-miR-500-3p on MIN6 cells are as follows Figure 13As shown, where A is the experimental result of the static insulin secretion stimulation experiment (GSIS), B is the experimental result of the CCK-8 experiment, C is the experimental result of the EdU experiment, and D is the quantitative result of the EdU experiment.

[0203] The static insulin secretion stimulation experiment (GSIS) indicated that overexpression of mmu-miR-500-3p could damage the insulin secretion of MIN6 cells under high glucose stimulation (N = 6). The CCK-8 experiment and the EdU experiment suggested that overexpression of its mouse homolog mmu-miR-500-3p significantly inhibited the proliferation, differentiation ability and viability of MIN6 cells (N = 9). The above results together indicated that hsa-miR-502-3p / mmu-miR-500-3p might participate in the pathological process of diabetes by having a dual effect on pancreatic islet β cells, affecting both the secretion function and inhibiting the proliferation and differentiation of β cells.

[0204] 2.7.3 hsa-miR-502-3p targets SUR1 to affect beta cell function

[0205] To clarify whether hsa-miR-502-3p plays the above functional role by targeting SUR1, this study first conducted a gain-of-function experiment, that is, transfected with hsa-miR-502-3p / mmu-miR-500-3p mimic, and then detected the expression level of SUR1. The results are as follows Figure 14 shown. Among them, A is the mRNA expression result of SUR1 detected by RT-qPCR in HEK-293T cells transfected with hsa-miR-502-3p mimic, B is the protein expression result of SUR1 detected by Western Blot in HEK-293T cells, C is the protein quantification result of SUR1 detected by Western Blot in HEK-293T cells; D is the mRNA expression result of SUR1 detected by RT-qPCR in MIN6 cells transfected with mmu-miR-500-3p mimic, E is the protein quantification result of SUR1 detected by Western Blot in MIN6 cells, and F is the protein quantification result of SUR1 detected by Western Blot in MIN6 cells.

[0206] It can be seen from the appendix Figure 14 that SUR1 was downregulated by about 75% at the mRNA level (in appendix Figure 14 A: in HEK-293T cells, P < 0.0001; in appendix Figure 14 D: in MIN6 cells, downregulated by about 50%, P < 0.0001) and at the protein level (in appendix Figure 14 B-C: in HEK-293T cells, downregulated by about 50%, P < 0.05; in appendix Figure 14In MIN6 cells, there was a significant inhibition of approximately 60% downregulation (P < 0.001).

[0207] To further exclude non-specific interference, in this study, a dual-luciferase reporter system was also used. The wild-type (WT) and mutant (MUT) sequences of the 3’UTR of SUR1 were respectively constructed into the reporter vector, and the results are shown in the appendix Figure 15 as follows. As can be seen from the appendix Figure 15 , hsa-miR-502-3p only specifically inhibited the luciferase activity of the wild-type vector (a decrease of approximately 40%, P < 0.0001), thus confirming their direct binding at the molecular interaction level. The above results together indicate that hsa-miR-502-3p exerts its effects on insulin secretion and pancreatic islet β-cell proliferation and differentiation by targeting and regulating SUR1.

[0208] In summary, in this study, we first identified hsa-miR-502-3p as a potential biomarker for long-term remission after SIIT through machine learning combined with small-RNA omics data. Subsequently, RT-qPCR was used for further verification in an independent cohort, and it was confirmed that low expression of baseline hsa-miR-502-3p was significantly associated with long-term remission after SIIT, and its predictive efficacy was independent of traditional clinical indicators. Validation across two multi-center RCT cohorts ensured the reliability of has-miR-502-3p as a biomarker. The further constructed combined model incorporated hsa-miR-502-3p with the functional parameter ISSI-2 of pancreatic islet β-cells, increasing the ROC-AUC to 83.7%, an 8.8% increase compared to the single ISSI-2 model, further improving the model's recognition ability and achieving a better balance between predictive efficacy and clinical interpretability. At the same time, the clinical decision curve of the combined model indicated that at a 40% risk threshold, the combined model could avoid 69% of unnecessary subsequent interventions for SIIT, significantly optimizing the allocation of medical resources. The online web-based prediction tool for long-term remission of SIIT developed based on the combined model achieved interactive visual risk assessment, providing technical support for precise intervention by medical institutions and clinicians.

[0209] At the molecular mechanism level, this study for the first time confirmed that hsa-miR-502-3p targets SUR1 / ABCC8, impairing insulin secretion and cell proliferation and differentiation of pancreatic islet β cells under high glucose stimulation. Before SIIT treatment, the higher the abundance of hsa-miR-502-3p in the serum of T2DM patients, the more severe the damage to SUR1 on the pancreatic islet β cell membrane. Under the dual stimulation of chronic metabolic stress (such as glucolipotoxicity and insulin resistance) and hsa-miR-502-3p, the calcium excitotoxicity caused by damaged SUR1 and the transdifferentiation or dedifferentiation of β cells will lead to irreversible permanent damage to pancreatic islet β cells. Even after SIIT relieves glucotoxicity, it is not easy for patients to restore normal pancreatic islet β cell function. Therefore, the amount of residual pancreatic islet β cell function that patients have at baseline is likely to be the key determinant of whether SIIT can reverse diabetes.

[0210] The results of this study have multiple implications for clinical practice. First, baseline hsa-miR-502-3p detection can be used as a clinical tool for making precise intervention decisions after SIIT. For patients with a higher baseline hsa-miR-502-3p expression level, they need to be advised to strengthen their own blood glucose detection and maintain closer follow-up observations to enhance the β cell protection effect of SIIT. Second, T2DM patients with high baseline hsa-miR-502-3p expression show more significant β cell function decline. Early identification and precise stratification of T2DM patients with different characteristics, and the development of small molecule inhibitors targeting the hsa-miR-502-3p / SUR1 pathway may become a new direction for diabetes reversal treatment and enhancing the efficacy of SIIT.

[0211] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and the descriptions in the specification only illustrate the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention will have various changes and improvements, and these changes and improvements all fall within the scope of the present invention claimed. The scope of protection claimed by the present invention is defined by the appended claims and their equivalents.

Claims

1. A marker for predicting the effect of insulin intensive therapy, characterized in that: The biomarker includes hsa-miR-502-3p in serum.

2. A marker for predicting the effect of insulin intensive therapy according to claim 1, characterized in that: The biomarker also includes ISSI-2.

3. A prediction model for predicting the effect of insulin intensive therapy, characterized in that: The prediction model is constructed based on the biomarker described in Claim 1 or 2.

4. The prediction model according to claim 3, wherein: The prediction model includes a multi-factor logistic regression equation model and a nomogram model.

5. The prediction model according to claim 4, wherein: The nomogram model includes: The first row is a score scale with a score range of 0 to 100; The second row is the weight before insulin intensive treatment, with a range of 50 - 110 kg; The third row is gender, including male and female; The fourth row is age, with a range of 20 - 70 years old; The fifth row is the body mass index BMI, with a range of 20 - 34 kg / m 2 ; The sixth row is glycated hemoglobin, with a range of 8% - 19%; The seventh row is insulin sensitivity index - 2, with a range of 50 - 650; The eighth row is hsa-miR-502-3P, with a range of -2.5 - 2.6; The ninth row is the prediction score, with a range of 280 - 420; The tenth row is the probability of maintaining remission after SIIT, with a range of 0% - 100%.

6. Application of a reagent for detecting hsa-miR-502-3p in the preparation of a product for predicting the effect of insulin intensive treatment for diabetes.

7. The application according to claim 6, characterized in that: The product includes a kit, a fluorescent probe, and a chip.

8. Application of a small molecule inhibitor of the hsa-miR-502-3p / SUR1 pathway in the preparation of a drug for treating diabetes.

Images